Everyone's shipping Agents, No one's measuring their impact.
‍

At Amplitude, we worked with teams building AI agents to understand how they monitored quality, investigated failures, and learned from real user interactions. We found that teams were piecing together information across logs, traces, and internal tools just to understand what happened.

AI product teams were flying blind after launch.
‍

Teams had plenty of data, but no clear way to turn it into actionable insight. Finding a problematic interaction meant manually digging through sessions, while understanding patterns across thousands of users was slow and difficult to scale.

‍

‍

What if teams could measure the ROI of every agent interaction?

We designed Agent Analytics to help teams move from individual sessions to patterns across their user base. Teams could identify where users were abandoning tasks, experiencing friction, or giving negative feedback, then drill into the interactions driving those outcomes.

‍

Everyone's shipping Agents, No one's measuring their business outcomes.
‍

At Amplitude, we worked with teams building AI agents to understand how they monitored quality, investigated failures, and learned from real user interactions. We found that teams were piecing together information across logs, traces, and internal tools just to understand what happened.

‍

‍

Turning millions of agent interactions into actionable product insights.
‍

I led product design and strategy from early concept through launch, working directly with 30+ design partners to shape the experience. We explored ways to combine behavioral metrics with AI-powered analysis so teams could quickly move from “what happened?” to “how often is this happening, who is affected, and what should we fix?”

‍

‍

The most valuable insights were the ones tied to user outcomes.
‍

We continuously tested whether teams could move from a high-level metric to the underlying interactions driving it. This pushed the product beyond dashboards and monitoring toward a workflow for understanding why key metrics were changing and where teams should focus their attention.

‍

‍